2025
Hey AI, Generate Me a Hardware Code! Agentic AI-based Hardware Design & Verification
An agent-based approach to hardware design and verification that combines AI agents with human-in-the-loop iteration to converge toward working verification flows.
Key contributions
- Iterative agentic workflow that self-corrects over multiple passes
- Evaluation across representative designs with strong coverage outcomes
- Emphasis on usability and configurability, not just a one-off demo
Formal that “Floats” High: Formal Verification of Floating Point Arithmetic
A scalable approach to floating-point verification using direct RTL-to-RTL model checking against a golden reference, supported by staged helper assertions and counterexample-guided refinement.
Key contributions
- RTL-to-RTL checking to reduce abstraction gaps and translation overhead
- Divide-and-conquer proof structure with modular stages and helper lemmas
- AI-assisted property generation with human refinement, plus coverage-driven analysis
2022
Thermal Alarm Handling in Safety Critical ECUs for Automated Vehicle Using AI and Machine Learning
Improving thermal monitoring robustness in safety-critical ECUs by incorporating real-world use-cases that are often missed during design-time validation, with a focus on reducing false triggering and validation cost.
Key contributions
- Highlights gaps between lab validation and real-world thermal events
- Motivates stronger characterization and scenario coverage for robust monitoring
- Practical impact: fewer false warnings and reduced validation/warranty cost
Related topics
Want the artifacts behind the papers?
If you want to discuss the workflows behind these results—formal proof structure, automation, or GenAI-assisted verification—reach out.